I was struck by the fact that Classic Gauge Theory in EM says that E & B fields are invariant under transformations of the potential, $V$, and vector potential, $A$, by any function, $f$, that is twice continuously differentiable function that depends on position and time. That is, \begin{align} \mathbf{A} &\mapsto \mathbf{A} + \nabla f\\ V &\mapsto V - \frac{\partial f}{\partial t} \end{align}

Which is remarkably similar to the invariance of action integral under the addition of the total derivative with respect to $t$ of any function of position and time.

I wanted to dig more into this, and I saw that gradient in spherical has the same terms as the square rooted the dr-dot squared portion of the Lagranian (ie the kinetic) and divided by the non-dotted term. It's hard to put what I mean into words. However, please refer to the comparison below.

Gradient: $$\boldsymbol{\nabla}{f} = \mathbf{e}_r\, \frac{\partial{f}}{\partial{r}} + \mathbf{e}_\theta\,\frac{1}{r } \,\frac{\partial{f}}{\partial{\theta}} + \mathbf{e}_\phi\,\frac{1}{r\,\sin{\theta}} \,\frac{\partial{f}}{\partial{\phi}}$$ Lagrangian: $$\vec v^2 = \dot r^2 + r^2 \dot \theta^2 + r^2 \sin^2(\theta) \dot \phi^2$$

Am I grasping at straws with a connection between Gauge invariance and Action invariance? Am I grasping at straws for connection between gradient and kinetic?


1 Answer 1


The equations of motion of the Lagrangian of electrodynamics are gauge invariant under the gauge transformation $A^\mu = A^\mu + \partial^u f$ where $A^0=V$.

You are seeing similarities in these expressions because of how vectors and basis vectors transform into spherical coordinates and/or from the metric of spherical coordinates, depending on how you are approaching the question.

Consider polar coordinates.

We want to express $v=\frac{dx}{dt} i + \frac{dy}{dt}j$ in polar coordinates.

$x=r \cos(\theta)$

$y=r \sin(\theta)$

Use the Chain Rule and Product Rule.

$\dot{x}= \dot{r} \cos{\theta} -r \sin(\theta) \dot{\theta}$

$\dot{y}= \dot{r} \sin{\theta} +r \cos(\theta) \dot{\theta}$

$v=(\dot{r} \cos{\theta} -r \sin(\theta) \dot{\theta})i + (\dot{r} \sin{\theta} +r \cos(\theta) \dot{\theta})j $

$\text{velocity}=(\dot{x},\dot{y})=\dot{x} \hat{i}+\dot{y} \hat{j}=f(r,\theta) \hat{r} + g(r,\theta) \hat{\theta}$

Use the fact that basis vectors transform like covariant tensors of rank 1.

$\hat{r}= \frac{dx}{dr} \hat{i} + \frac{dy}{dr} \hat{j} \\$

$\hat{\theta}= \frac{dx}{d\theta} \hat{i}+\frac{dy}{d\theta} \hat{j} $


$\hat{r} \cdot \text{velocity} = f(r,\theta)$

$\hat{\theta} \cdot \text{velocity} = g(r,\theta)$.


$f(r,\theta)=\cos(\theta)[(\dot{r} \cos{\theta} -r \sin(\theta) \dot{\theta}]+ \sin(\theta)[(\dot{r} \sin{\theta} +r \cos(\theta) \dot{\theta})] = \dot{r}$

(using $\cos^2 (\theta) + \sin^2 (\theta)=1)$

The $\theta$ component of velocity can be found similarly.

This ultimately comes from the fact that distance is an invariant such that

$dx^2 + dy^2 =(\frac{dx}{dr} \frac{dx}{dr} +\frac{dy}{dr} \frac{dy}{dr})dr^2 +(\frac{dx}{d\theta} \frac{dx}{d\theta} +\frac{dy}{d\theta} \frac{dy}{d\theta})d\theta^2 = \hat{r} \cdot \hat{r} \space dr^2 + \hat{\theta} \cdot \hat{\theta} d\theta^2 $

The mathematics for the gradient is similar. The gradient is a covariant vector and the velocity vector here is a contravariant vector. These transform "oppositely" to each other and that's why things are moving around from the denominator to the numerator.

  • $\begingroup$ Thank you! I'll return to think when I've got better understanding of tensor calc which we just started in EM $\endgroup$
    – EEH
    Oct 28, 2022 at 20:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.